JP2008112056A - Audio sigmal processor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve the quality of reproduced sound of a digital audio sigmal. <P>SOLUTION: A minimal and maximal point detector 1 detects the local minimal point and the local maximal point in a waveform of an input digital audio sigmal, and a tilt calculation section 2 calculates a tilt value of a straight line for connecting adjoining the local minimal point and the local maximal point. A normalization value calculation section 3 calculates a value in which the value on the straight line for connecting the adjoining the local minimal point and the local maximal point, is subtracted from each sample value, as a normalization value. An auxiliary value calculation section 4 calculates the value in which the normalization value for each sample calculated in the normalization value calculation section 3, is added to the value in which a coefficient according to the sampling interval number between the adjacent local minimal point and the local maximal point is multiplied with the normalization value, as the auxiliary value. An output data creation section 6 adds the auxiliary value for each sample which is calculated in the auxiliary value calculation section 4 to the value on the straight line for connecting the adjoining the local minimal point and the local maximal point at timing of each sample, and outputs it. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an audio signal processing apparatus that improves the quality of reproduced sound of a digital audio signal.

  In general, high-frequency components are removed by conversion from an analog audio signal to a digital audio signal (including compressed audio data). For this reason, even if a digital audio signal is directly D / A converted and reproduced by a speaker, the original sound cannot be reproduced faithfully.

Therefore, techniques as described in Patent Documents 1 and 2 have been proposed in order to bring a digital audio signal closer to the original sound. In Patent Documents 1 and 2, assuming that the original waveform of the digital audio signal on the time axis (the waveform of the digital audio signal when the high frequency component is not removed) is close to a rectangular wave, the input data waveform is changed to a rectangular wave. It is corrected so that it is closer. In this method, as shown in FIG. 5, with respect to samples before and after the sample having the minimum value in the input data waveform, a predetermined value is subtracted from each sample value, and the samples before and after the sample having the maximum value are subtracted. On the other hand, the input data waveform is approximated to a rectangular wave by adding a predetermined value to each sample value.
Japanese Patent Laid-Open No. 2002-169597 JP 2002-189498 A

  By the way, when the original waveform is an acute waveform such as a triangular wave, the waveform of the digital audio signal is close to a waveform such as a rectangular wave due to loss of high frequency components by A / D conversion. ing. When such a digital audio signal is corrected by the techniques described in Patent Documents 1 and 2, the waveform is corrected so as to be close to a rectangular wave, so that it is away from the original acute waveform. Become.

  As described above, in the techniques described in Patent Documents 1 and 2, the input data waveform cannot be brought close to the original waveform as in the case where the original waveform is an acute waveform, and the sound quality improvement effect cannot be obtained. There was a thing.

  The present invention has been made in view of the above. In the sound quality improvement processing of a digital audio signal, even if the original waveform of the digital audio signal is a waveform such as a rectangular wave or an acute waveform, An object of the present invention is to provide an audio signal processing apparatus capable of improving the sound quality of the sound.

  In order to achieve the above object, the audio signal processing apparatus of the present invention compares the sample value of the sample with the sample value of the immediately previous sample for each sample of the digital audio signal input from the outside, Based on the comparison result, the extreme point detecting means for detecting the minimum point and the maximum point in the waveform of the digital audio signal, and the slope for calculating the slope value of the straight line connecting the adjacent minimum point and the maximum point in the waveform A value indicating how much the sample value of each sample between the adjacent minimum point and the maximum point is deviated from the straight line using the calculation means and the inclination value calculated by the inclination calculation means. According to the normalized value calculating means for calculating the normalized value, the normalized value calculated by the normalized value calculating means, and the number of sampling intervals between the adjacent minimum and maximum points. Correction value calculation means for calculating a correction value for correcting each sample between the adjacent minimum point and the maximum point in a direction in which the deviation from the straight line increases, using a number; Output data generating means for adding the correction value for each sample calculated by the correction value calculating means to the value on the straight line at the timing of each sample and outputting the added value.

  Also, the audio signal processing apparatus of the present invention compares the sample value of the sample with the sample value of the immediately previous sample for each sample of the digital audio signal input from the outside, and based on the comparison result A maximum point detecting means for detecting a minimum point and a maximum point in the waveform of the digital audio signal; a slope calculating means for calculating a slope value of a straight line connecting adjacent minimum points and maximum points in the waveform; and Using the inclination value calculated by the inclination calculating means, the value on the straight line at the timing of each sample between the adjacent local minimum point and local maximum point is calculated, respectively, and the calculation result is the sample of each corresponding sample. A normalized value calculating means for calculating a value subtracted from the value as a normalized value for each sample, and the positive value for each sample calculated by the normalized value calculating means. A correction value for calculating a correction value for each sample by adding a normalized value and a value obtained by multiplying the normalized value by a coefficient corresponding to the number of sampling intervals between the adjacent minimum and maximum points It is characterized by comprising: a calculating means; and an output data generating means for adding the correction value for each sample calculated by the correction value calculating means to the value on the straight line at the timing of each sample and outputting it.

  According to the audio signal processing apparatus of the present invention, each sample between adjacent local minimum points and local maximum points in the waveform of a digital audio signal is greatly displaced from a straight line connecting the adjacent local minimum points and local maximum points. Since the waveform is corrected in the direction, it can be corrected so that the waveform of the digital audio signal is close to the original waveform, regardless of whether the original waveform is a rectangular waveform or an acute waveform. The sound quality of the reproduced sound of the signal can be improved.

  The best mode for carrying out an audio signal processing apparatus of the present invention will be described below with reference to the drawings.

  FIG. 1 is a block diagram showing a configuration of an audio signal processing apparatus according to an embodiment of the present invention. As shown in FIG. 1, the audio signal processing apparatus according to the present embodiment includes an extreme point detection unit 1 that detects a local minimum point and a local maximum point in a waveform of a digital audio signal input from the outside, and a waveform of the digital audio signal. The slope calculation unit 2 that calculates the slope value of a straight line connecting the local minimum point and the local maximum point, and each sample between the local minimum point and the local maximum point using the slope value calculated by the slope calculation unit 2 Is a normalization value calculation unit 3 that calculates a normalization value that is a value indicating how much is deviated with respect to a straight line connecting adjacent local minimum points and local maximum points, and a normalization value calculated by the normalization value calculation unit 3 Each of the samples between the adjacent local minimum points and the local maximum points, and the coefficient corresponding to the number of sampling intervals between the adjacent local minimum points and the local maximum points. Correct in a direction that increases the deviation from the connecting straight line. A correction value calculation unit 4 that calculates a correction value for each sample, a coefficient table 5 that stores a coefficient corresponding to the number of sampling intervals between adjacent local minimum points, and a correction value calculation unit 4 And an output data generation unit 6 that adds the correction value for each sample to a value on a straight line connecting adjacent local minimum points and local maximum points at the timing of each sample.

  Next, the operation of the audio signal processing apparatus according to this embodiment will be described.

  When a digital audio signal is input from the outside, the pole detection unit 1 temporarily records several samples of the input digital audio signal. Then, for each recorded sample, the sample value of the sample is compared with the sample value of the previous sample in time to detect whether it is large, small, or the same. Based on the comparison result, the pole detection unit 1 detects a minimum point and a maximum point in the waveform of the input digital audio signal.

  Next, the slope calculation unit 2 is adjacent in the waveform of the digital audio signal based on the local minimum point detected by the local point detection unit 1, the value of the local maximum point, and the number of sampling intervals between the local minimum point and the local maximum point. The slope value of the straight line connecting the minimum point and the maximum point to be calculated is calculated.

  The normalized value calculation unit 3 uses the inclination value calculated by the inclination calculation unit 2 on the straight line connecting the adjacent minimum point and the maximum point at the timing of each sample between the adjacent minimum point and the maximum point. The value obtained by subtracting the calculated result from the sample value of each corresponding sample is calculated as a normalized value for each sample.

  Then, the correction value calculation unit 4 uses the normalized value for each sample calculated by the normalized value calculation unit 3 and a coefficient corresponding to the number of sampling intervals between the adjacent minimum points and maximum points as the normalized value. A value obtained by adding the multiplied values is calculated as a correction value for each sample. Here, a coefficient corresponding to the number of sampling intervals between adjacent local minimum points and local maximum points is acquired from the coefficient table 5.

  An example of the coefficient table 5 is shown in FIG. As shown in FIG. 2, in the coefficient table 5, a relatively large coefficient α = 1/2 is set when the number of sampling intervals between adjacent minimum points and maximum points is small (the number of sampling intervals = 2 to 5). As the number of sampling intervals increases, it is set to decrease stepwise.

  Then, the output data generation unit 6 adds the correction value for each sample calculated by the correction value calculation unit 4 to a value on a straight line connecting adjacent local minimum points and local maximum points at the timing of each sample. Output.

  A case where a digital audio signal having a time axis waveform as shown in FIG. 3A is input will be described as a first example of a processing target when digital audio signal correction processing is performed as described above.

When the digital audio signal shown in FIG. 3A is input, the pole detection unit 1 detects the sample A ₃ as a minimum point and the sample A ₈ as a maximum point. Here, the sample values of the samples A _{3 to} A ₈ are a _{3 to} a _8, and a ₃ = 100, a ₄ = 200, a ₅ = 460, a ₆ = 680, a ₇ = 880, a ₈ = 1000 To do.

Next, the inclination calculation unit 2 calculates the inclination value of the straight line 11 connecting the sample A ₃ and the sample A ₈ . Since the number of sampling intervals between the sample A ₃ and the sample A ₈ is 5, the slope value k ₁ is calculated as (Equation 1) below.

_{k 1 = (1000-100) / 5} = 180 ( Equation 1)

Next, the normalized value calculation unit 3 calculates normalized values n _{3 to} n ₈ for the samples A _{3 to} A ₈ . The normalized values n _{3 to} n ₈ are values that serve as indicators of how much the samples A _{3 to} A ₈ are shifted from the straight line 11 (see FIG. 3B). The normalized values n _{3 to} n ₈ are obtained by changing the values on the straight line 11 at the respective timings of the samples A _{3 to} A ₈ from the sample values a _{3 to} a _{8 as in} the following (Formula 2) to (Formula 7). Calculated by subtraction.

n ₃ = 100-100 = 0 (Equation 2)
n ₄ = 200− (100 + 180 × 1) = − 80 (Formula 3)
n ₅ = 460− (100 + 180 × 2) = 0 (Formula 4)
n ₆ = 680− (100 + 180 × 3) = 40 (Formula 5)
n ₇ = 880− (100 + 180 × 4) = 60 (Formula 6)
n ₈ = 1000−1000 = 0 (Formula 7)

Then, the correction value calculating unit 4, the correction value _p 3 ~p ₈ for samples _A 3 to A ₈ are calculated. The correction values p _{3 to} p ₈ are sampled between the normalized values n _{3 to} n ₈ for each sample and the sample A ₃ and the sample A ₈ as in the following (formula 8) to (formula 13). It is calculated by adding a value obtained by multiplying the normalized value n _{3 to} n ₈ by a coefficient α corresponding to the number of intervals. Here, since the number of sampling intervals between the sample A ₃ and the sample A ₈ is 5, the coefficient α is ½ from FIG.

p ₃ = 0 (Formula 8)
p ₄ = −80 + (− 80 × 1/2) = − 120 (Equation 9)
p ₅ = 0 (Formula 10)
p ₆ = 40 + (40 × 1/2) = 60 (Formula 11)
p ₇ = 60 + (60 × 1/2) = 90 (Equation 12)
p ₈ = 0 (Formula 13)

Next, the output data generation unit 6 calculates the corrected sample values a ₃ ′ to a ₈ ′. The corrected sample values a ₃ ′ to a ₈ ′ are obtained by changing the correction values p _{3 to} p ₈ for each sample on the straight line 11 at the timing of each sample, as in the following (Formula 14) to (Formula 19). It is calculated by adding to the value of.

a ₃ ′ = 100 + 0 = 100 (Formula 14)
a ₄ ′ = (100 + 180 × 1) −120 = 160 (Formula 15)
a ₅ ′ = (100 + 180 × 2) + 0 = 460 (Formula 16)
a ₆ ′ = (100 + 180 × 3) + 60 = 700 (Formula 17)
a ₇ ′ = (100 + 180 × 4) + 90 = 910 (Formula 18)
a ₈ ′ = 1000 + 0 = 1000 (Formula 19)

The waveform after the correction process of the first example is shown in FIG. In FIG. 3C, samples A ₄ ′, A ₆ ′, and A ₇ ′ indicate sample points after correction processing corresponding to the samples A ₄ , A ₆ , and A ₇ , respectively, and the waveforms before the correction processing are dotted lines The waveform after the correction processing is indicated by a solid line.

In the first example, between samples A ₃ to A ₈ of the waveform of the input digital audio signals, since the slope of the waveform at sample A _3, A ₈ is smaller than the slope of the straight line 11, the original waveform rectangular It is expected to be a wave-like waveform. Therefore, by performing the correction process described above, as shown in FIG. 3C, the slope of the waveform after correction in samples A ₃ and A ₈ is made smaller than the slope of the waveform before correction. Thus, it can be approximated to a waveform like an expected rectangular wave, thereby improving the sound quality of the reproduced sound.

  Next, a case where a digital audio signal having a time axis waveform as shown in FIG. 4A is input as a second example of the correction processing target will be described.

When the digital audio signal shown in FIG. 4 (a) is input, the sample B ₃ is detected as the minimum point by the pole detection unit 1, the sample B ₈ is detected as the maximum point. Here, the sample values of the samples B _{3 to} B ₈ are b _{3 to} b ₈ and b ₃ = 100, b ₄ = 330, b ₅ = 500, b ₆ = 640, b ₇ = 780, b ₈ = 1000. To do.

Next, the inclination calculation unit 2 calculates the inclination value of the straight line 12 connecting the sample B ₃ and the sample B ₈ . Since the number of sampling intervals between the sample B ₃ and the sample B ₈ is 5, the slope value k ₂ is calculated as (Equation 20) below.

_{k 2 = (1000-100) / 5} = 180 ( Equation 20)

Next, in the normalized value calculation unit 3, the normalized values m _{3 to} m for the samples B _{3 to} B ₈ are obtained in the same manner as in the first example, as in the following (Expression 20) to (Expression 25). ₈ is calculated.

m ₃ = 100-100 = 0 (Formula 20)
m ₄ = 330− (100 + 180 × 1) = 50 (Formula 21)
m ₅ = 500− (100 + 180 × 2) = 40 (Formula 22)
m ₆ = 640− (100 + 180 × 3) = 0 (Formula 23)
m ₇ = 780− (100 + 180 × 4) = − 40 (Equation 24)
m ₈ = 1000−1000 = 0 (Formula 25)

Then, the correction value calculating unit 4, as shown in the following (Equation 26) to (Equation 31), in the first example and the same method, the correction value _q 3 to q ₈ for the sample _B 3 .about.B ₈ is Calculated.

q ₃ = 0 (Formula 26)
q ₄ = 50 + (50 × 1/2) = 75 (Equation 27)
q ₅ = 40 + (40 × 1/2) = 60 (Equation 28)
q ₆ = 0 (Equation 29)
q ₇ = −40 + (− 40 × 1/2) = − 60 (Equation 30)
q ₈ = 0 (Formula 31)

Next, the output data generation unit 6 calculates the corrected sample values b ₃ ′ to b ₈ ′ by the same method as in the first example, as in the following (Equation 32) to (Equation 37). .

b ₃ ′ = 100 + 0 = 100 (Formula 32)
b ₄ ′ = (100 + 180 × 1) + 75 = 355 (Formula 33)
b ₅ ′ = (100 + 180 × 2) + 60 = 520 (Formula 34)
b ₆ ′ = (100 + 180 × 3) + 0 = 640 (Formula 35)
b ₇ ′ = (100 + 180 × 4) −60 = 760 (Formula 36)
b ₈ ′ = 1000 + 0 = 1000 (Formula 37)

The waveform after the correction process of the second example is shown in FIG. In FIG. 4C, samples B ₄ ′, B ₅ ′, and B ₇ ′ indicate sample points after correction processing corresponding to the samples B ₄ , B ₅ , and B ₇ , respectively, and the waveforms before the correction processing are dotted lines The waveform after the correction processing is indicated by a solid line.

In a second example, between samples B ₃ .about.B ₈ waveforms of the input digital audio signals, since the slope of the waveform at sample B _3, B ₈ is greater than the slope of the straight line 12, the original waveform acute It is expected to be a typical waveform. Therefore, by performing the correction process described above, as shown in FIG. 4C, the slope of the waveform after correction in samples B ₃ and B ₈ is made larger than the slope of the waveform before correction. Thus, it can be close to the expected sharp waveform, and thereby the quality of the reproduced sound can be improved.

  As described above, according to the present embodiment, each sample between adjacent local minimum points and local maximum points in the waveform of the digital audio signal is greatly deviated from the straight line connecting the adjacent local minimum points and local maximum points. Correct in the direction. For this reason, even if the original waveform is a rectangular waveform like the first example, or the original waveform is an acute waveform like the second example, the digital audio The signal waveform can be corrected so as to be close to the original waveform, and the quality of the reproduced sound can be improved.

  The functions of the audio signal processing apparatus described above may be realized by a computer by a program. This program may be read from a recording medium and loaded into a computer, or may be transmitted via a communication network and loaded into a computer.

It is a block diagram which shows the structure of the audio | voice signal processing apparatus which concerns on embodiment of this invention. It is a figure which shows an example of a coefficient table. It is a figure for demonstrating the 1st example of the waveform correction process by the audio | voice signal processing apparatus shown in FIG. It is a figure for demonstrating the 2nd example of the waveform correction process by the audio | voice signal processing apparatus shown in FIG. It is explanatory drawing which shows the processing waveform by a prior art.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Pole detection part 2 Inclination calculation part 3 Normalization value calculation part 4 Correction value calculation part 5 Coefficient table 6 Output data generation part

Claims (2)

For each sample of the digital audio signal input from the outside, the sample value of the sample is compared with the sample value of the previous sample in time, and based on the comparison result, the minimum point in the waveform of the digital audio signal And extreme point detection means for detecting the local maximum point,
An inclination calculating means for calculating an inclination value of a straight line connecting a local minimum point and a local maximum point adjacent in the waveform;
Normalization that is a value indicating how much the sample value of each sample between the adjacent minimum point and the maximum point is deviated from the straight line by using the inclination value calculated by the inclination calculating means. A normalized value calculating means for calculating a value;
Using the normalized value calculated by the normalized value calculation means and a coefficient corresponding to the number of sampling intervals between the adjacent minimum point and the maximum point, between the adjacent minimum point and the maximum point Correction value calculation means for calculating a correction value for correcting each sample in a direction in which the deviation from the straight line increases,
An audio signal processing apparatus comprising: output data generating means for adding the correction value for each sample calculated by the correction value calculating means to the value on the straight line at the timing of each sample and outputting the result. For each sample of the digital audio signal input from the outside, the sample value of the sample is compared with the sample value of the previous sample in time, and based on the comparison result, the minimum point in the waveform of the digital audio signal And extreme point detection means for detecting the local maximum point,
An inclination calculating means for calculating an inclination value of a straight line connecting a local minimum point and a local maximum point adjacent in the waveform;
Using the inclination value calculated by the inclination calculating means, the value on the straight line at the timing of each sample between the adjacent minimum point and the maximum point is calculated, and the calculation result is calculated for each corresponding sample. Normalized value calculation means for calculating a value subtracted from the sample value as a normalized value for each sample;
The normalized value for each sample calculated by the normalized value calculating means is added to the value obtained by multiplying the normalized value by a coefficient corresponding to the number of sampling intervals between the adjacent minimum and maximum points. Correction value calculating means for calculating the value obtained as a correction value for each sample;
An audio signal processing apparatus comprising: output data generating means for adding the correction value for each sample calculated by the correction value calculating means to the value on the straight line at the timing of each sample and outputting the result.

Images